Simple question about the logical shape of a velocity curve

The exponential curve is the most logical one to start with.
Why? Ok, it is easy to agree that 0 = 0 and 127 = 127 if you want to use the complete range. In the exponential form 127 would be 1 but this is just a factor.
Now you have one point for the ppp and one for the fff and both are fixed.
Now think of what leads to a readout of velocity values in the keyboard. Usually there are a few switches, which are triggered at certain points of the key travel or hammer travel. The time passing between the switch triggers is transfered to the velocity. It is a continous transfer function, in fact most manufacturers don't care too much about it. So the function between the real key velocity and the MIDI velocity is likely going to be of a low order.
In an ideal world the middle of the ppp and the fff would be mf. The middle means a straight line.
Now if the world happens to be not so ideal and the order of the function is not 0 or your taste just demands it the middle is too loud or too soft. Lets say it is 10% off, that means it is not 50% but 45%.
What do you think how much will the middle between 0 and your new mf point be off? Why should the function in that part of the velocity curve be different to any other part? The best guess is if the overall middle has been 10% off, the quarter is 10% off too. The arithmetic center (straight line) would be (0+45)/2 = 22,5. 10% off means 22,5*0,90 = 20,25. That is exactly what an exponential curve does.

If the amount of deviation from the straight line is very small, the exponential function and a circle segment are looking very similar. But imagine the most extreme circle segment you can use. It would be the quarter of a cirle with its center at one corner of the entry box. You can easily see that the amount of "offness" in the parts of the function are very different. And you are limited by the quarter circle, if you want to go further the circle segment will go outside of the box.

So the exponential function is really the best one to start with but if it comes to your taste and desired feeling of the action there may be more sections in your velocity curve. Sometimes also the keyboard manufacturer has really done some more math, e.g. to make the soft notes easier to play and this doesn't conform to your habits. E.g. you can end up with a S or a ?. Then actually you have two or three sections of some exponential curves blended together.

That's interesting, I guess my playing style is different. I prefer the 'Slow Keyboard' preset for the FP-30/PHA-4, it opens up the action the way I like (see image below). But, like you, I also found the default curve was good enough (after adjusting it to hell and back).

Thanks, this is very useful too, I'll try and see where it gets me

Many things affect the shape of the vel curve; the straight line was ok for my Roland FP50, but on my Yamaha, an easy gradient bottom left gives me a nice pp range.  I also use the piano vel curve on lightest touch, so the PTq curve has to accommodate that too.
And if you change the dynamics, well guess the rest.  It's juggling and you have to learn how to do it.  I think if there was an easier way, the good guys at PTq would've employed it.
Just my thoughts.

I’ll have to try the dynamics at 50 db for my live setup - along with the ‘Player’ presets, which are also good for live performances.

When I have changed the PTQ dynamics I use 100 db because it gives me so much control. It feels like cruise control at different dynamic levels - but then at 100 db it’s a guilty pleasure and feels like cheating since it’s difficult to have that on an acoustic.

If I can just take a guess, I would maybe guess Philippe Guillaume wants to play with your mind or insure your playing (experience) with his software is a humbling one that after people rant and rave to others on how good the software is in their view or even about its presumed failure to sound genuinely like an actual acoustic to many of those you finally over the course of several years maybe even a decade arrive at some humility enough (that is) for you to admit simply to your own error and other mistakes only you yourself made about it  —via default settings of course.

The very same seems true of your presets labeled by the fxp extension that they got.  Which I see as a reference to MIDI XP the high resolution format and also an indication that maybe alludes to the fact with them you’ve only f’ed up somehow.  That’s the reality you really have!

I speak of course from experience.  My own great granddad who at first appeared on the surface possibly a simple farmer at the plot, had his say about institutions and any society by and large its own government.  He to me was full of convictions, no bull.  Also I lived on a ranch and throughout other communal holdings where the elders from macramé and whittle spoke to the problems the youth there were facing.  So consequently, I am very familiar sometimes with a lot of witty types.  (Smile.)

I must confess I haven't started looking into note-off velocity yet. Too busy trying to match note-on velocities with various manufacturers' products—or rather the other way around. I do think I once tried to find out if the hammer made more noise as I withdrew my finger faster than normal and I'm not sure if I remember this correctly but it seems like it did. On a PHA-IV—namely, an RD-88. Not sure if my perception was accurate, or distorted by the psychological impact of my hand's moving faster. Either way, I'm definitely going to try and reduce note-off velocity to see how it impacts the piano sound. Also, while I'm here, have you ever played on the previous generation of Roland's keyboards—PHA-III, which was called "Ivory Feel G"? I got my hands on a Juno DS-88 this week, which has a PHA-III, and somehow I find that keyboard to be more realistic than the PHA-IV—the keys are heavier yet softer, and the escapement is less noticeable, which make it closer to an acoustic... Is it possible that Roland Corporation's been cutting the costs and making worse keybeds than their previous generation?

Slightly off-topic here on note-off (again). For anyone not hearing the note-off, with a keyboard regardless of it having the note-off feature, lower the horizontal line in note-off to the bottom so that 0 value is achieved, and you will hear that the sound remains after releasing the key. Try then moving the horizontal line to 10 for example, then try 127 horizontal line, this will give the fastest key release.

Back on-topic regarding note-on, I use the default (normal) curve. I found the fast curve too dull, although in theory I wanted to play quietly, it sounded as though my keyboard was not triggering correctly.

Nick

Adjusting the velocity curve is very subjective if you don't have the real piano at hand. One way to control this curve would be for pianoteq to provide us with the theoretical velocity for its pianos as a function of the velocity of pressing the piano key or the time the key is pressed (10mm stroke). For example: :
0.25 s for a midi velocity of 20
0.15 s for a velocity midi of 40
0.08 s for a velocity midi of 100
This would allow you to compare to your own keyboard and adjust the curve the closest to the original instrument.
Pianoteq, do you have these values please ?
One could even dream of having the speed/velocity ratios for each of the 88 keys of the original piano. But that's a lot to ask.

Suggestion sent.

Basically this is a good idea, but for being able to reproduce the real piano, you would need additional information like the SPL at that given velocity.
Sounds like one more parameter? Unfortunately not. You need to specify where the SPL is measured, with which weighting and method and so on.
Maybe that is more something for Pianoteq 8: a virtual SPL meter or SPL readout at the mic positions.

After another few days of testing multiple velocity curves and multiple settings, I must say there's something really nice and realistic about playing the New-York Steinway Model D (Ptq 7) in wide-player mode, after cropping both ends of the curve by 8, on a Roland PHA-IV (namely, an RD-88) with default velocity curve. No need to curve the curve, I'm talking about a straight line from 8 to 120, as the PHA-IV basically hardly gets down to less than 8 or goes up to more than 120. On top of feeling natural (quite more so than with an exponential curve), it also has the advantage of not distorting the natural response of the keyboard too much: when you play with an exponential curve and, for some reason, get rid of it after recording, for instance in order to use another virtual instrument, the resulting velocities often don't make sense, which is much less the case if you linearly re-open the range from 8-120 to 0-128. Another possible thing to do is to crop 12 at the bottom and 8 at the top, which makes PP's a little harder to achieve and so the whole feeling closer to a real piano, though that means slightly moving away from the keyboard's natural response, but is still more natural than an exponential curve. All that being, of course, in my humble opinion.

Just like to say I love this technique which is excellent advice, thanks dazric:

While on my mind - I often like to perform (with a view to exporting to audio) using a player type preset, and then after the MIDI is done, select a recording type preset and make edits to bring it to its best final form. Apart from all other editable things, this usually includes velocity curve tweaks (if just wanting a little more or less tonal effect in different velocities). Saving these curves gives me a nice library of curves to swap in and out if I'm looking for the right response from my dpaino for a given piano/preset combo suited to a desired outcome.

After monkeying with dpianos so long, I really now don't believe there's really any perfect correct curve per se. We all have different personal equations for mf or fff etc. - and this alters with repertoire and so on - but maybe because I see things from an audio perspective more than a pure pianist-only POV - but hope it's useful to throw this in for food for thought. To me, without question, each piano/preset makes me want to "feel" it differently, under my fingers. I like some latency too - this goes hand in hand with what we 'feel' as much as how it helps us get into a comfortable creative stride.

Hi.

This is a very interesting thread, I have a sort of a velocity issue myself that I posted in a new thread but maybe this is a good place to search som answers:)   I'm looking for a way to adjust some kind of "global velocity balance" where it would be possible balance how easy it is to reach the higher/lower velocity levels in Pianoteq in different regions of the keyed. I'm not searching for a note-per-note edit, just a simple way, a slider or a wheel, where you can balance this. On my controller, the highest velocity level I can reach is about 120 in the treble and around 110 in the bass. It would be nice to be able to balance this in a simple way. Maybe Pianoteq has this function built in that I don't know about? Instead I can for example adjust the velocity curve for the lower/mid region, which can give a little harsh treble, and then compensate with adjusting hammer hardness/volume/dynamics in the note-per-note edit. This works fine but it would be nice with a simple balance slider. Does anyone have any suggestions?

VelPro a Springbeats product comes recommended by julien, one of the administrators: https://forum.modartt.com/viewtopic.php...55#p956855.

It permits individual key velocities.

I suggest visit the Springbeats website: https://springbeats.com/velpro/.

Hi Amen Ptah Ra

That looks like a great product that might solve my problem. I will try it out as soon as I can, thank you!